Introduction and Overview

Sunbird currently supports Question Sets and Videos separately. However, we have learned that engagement with videos is much higher if questions can be placed inside videos. Once the videos are made interactive, the learning experience is much more enjoyable for the end user.

With support for Interactive Videos in Sunbird, we will have the capability to:

1. Make any video interactive: The goal is to design this in such a way what any available video can be made interactive by adding any question set at any timestamp in the video.

2. Publish interactive videos as a new form of content: This will enable teachers to leverage interactive videos to improve the experience for learners

3. Analyse interactions at a deeper level: The interactive video platform will provide an exhaust for telemetry and analysis which will allow data about engagement at a very detailed level.

Proposal

To enable interactive videos on Sunbird, we will leverage as many existing components as possible to enable quick and efficient development. Currently, we envision these main workstreams:

• Asset Model: An asset model will be enhanced to support interception points in content. We will start with videos, but this work will be done keeping in mind that in the future, we will extend this concept to support interactive content of any kind. In fact, we could, in the future, extend the concept even to more than two assets. For example, a child could read a few pages in a document (PDF), watch a quick video to visually understand the concept, and quickly practice what she learnt, all in sequence.

• Creator: A new content creation UI will be created to enable any user of Sunbird to create interactive videos. They will pick from any preexisting video, and be able to add any existing question sets at any timestamp in the video. A possibility here is to allow for creation of new question sets inside the creation process.

• Player: We do not need a new consumption experience for interactive videos. The current video player can be enhanced to support the interactive video content models

• Telemetry and Analysis: Detailed interaction events will be captured when an interactive video is consumed. This will include, but is not limited to video playback quality, scrubbing (Pause, Rewind and Fast Forward) behaviour, question events (answering, skipping, etc) and score related data.

Requirement Specifications

Here are the use cases we are currently targeting

• Interactive Video Creation: As a creator, I need a simple interface with good User Experience to be able to create interactive videos.

• Interactive Video Consumption: As a learner/viewer, I would like to be able to view interactive videos similar to any other content type on Sunbird. When I log in and navigate to my available content, I should be able to find interactive videos easily. When I choose an interactive video, the experience should be engaging and rewarding.

• Associated user stories 

• Non-functional requirements

• Localization requirements  

• Telemetry requirements

• Dependencies

• Impact on other products

• Impact on existing data  

Interactive Video Creation

The process to create interactive videos should be smooth and seamless. For V1, users cannot upload interactive videos, but can only use existing videos to enhance them with question sets for interactivity

Interactive Video Consumption

This scenario describes the MVP for a user playing an interactive video. The expectation here is that users can find and consume interactive videos similar to any other sort of content on Sunbird.

Creator Dashboard

I believe we should also have a section on “Creator Dashboard” where creator can see metrics for their interactive videos, generate insights about what to improve.

Exception Scenarios

Wireframes

Creator Interface V1

• On uploading video content on the Sunbird platform, the user will get an option to add questions to the video

  

• On clicking <Add a new question>, the browse question dialog box will pop up from where the user can select a question/multiple questions from the existing question created

• The question will be added at the same timestamp where the user had clicked on the <Add a new question> button. They can edit the timestamp in the bottom left corner of the dialog box

  

• Users can browse through the available questions and question sets to add to the video. They will also have the option to use multiple filters -

  • Board

  • Class

  • Subject

  • Question Type (MCQ/Subjective)

  • Chapter

  • All Questions / My Questions

• Once they select the question set/s they want to add, they can configure their question set (which questions to show {check box} and which order {drag and change position} ) in the Configure your Question Set part on the right side of the dialog box

• Once the user is done selecting the question set/s, they can click on <Add Question Set> to add it to the video. If the user selects more than one existing question set, a new unique question set will be created at the backend to impose on the video.

  

• The added question/s will appear on the right side of the player with the timestamp and ability to edit (will open the question browsing dialog box) and delete the question

• A blue marker will be visible on the progress bar of the player at the timestamp where the question is added

• User can also drag and move the marker on the progress bar to change the timestamp of the question

• If the user adds two separate question sets at the same time stamps, it will be considered as one question set at the backend

  

• On clicking the question on the right side of the player, a preview of the question will be visible to the user

  

• User can submit the interactive video for review once they have added all desired questions

Player Interface V1

• Timestamps with questions overlayed will be bookmarked with a green marker

  

• The viewer will have the option to skip and revise a question set

  • <Skip> - This will allow you to bypass the question and proceed with the video

  • <Revise> - This will start playing the video from the last bookmarked timestamp. In case the first question is revised, the video will start from the beginning

• On submitting the answer, the viewer will have to click on <Proceed> for continuing with the video

  

Scenarios

Scenario 1: Multiple question sets at the same timestamp.

Following ways in which different question sets can get attached to the same timestamp:

Given: Creator added only one question at timestamp x (assumption even this one question is stored as question set internally).

When: Creator tries to ‘Add question’ at the same timestamp.

Then: Show previously selected question and allow user to add more & configure settings.

Given: There is a question set (even if it’s singleton set) at timestamp x

When: Creator tries to ‘Add question’ at a different timestamp but modifies the time to a point where question set is already present

Then: As soon as user modifies the time, reload the page such that questions from question set already present at the timestamp are shown as selected + any additional questions user might have selected, in the current Add Question flow before modifying the time. will also show as selected. (Essentially merge selection - questions already present in the set + new ones selected this time when ‘Add Question’)

Alternative: Do not allow user to modify time from ‘Add Question’ flow. User should navigate the video to the desired timestamp and initiate Add question from there.

Given: There is a question set at timestamp x and y

When: User select the question set at y (blue dot in the player or from the right pane showing list of question sets) and modifies the position of the timestamp by dragging blue dot to x.

Then: Show merged question set - the one already present at x + the questions from the one at y. User can cancel this or Save the question set. Cancel will mean, user will have two question sets - one at x, one at y - from where s/he had started.

For Future Release

1. In future releases, we will aim to enhance all types of content with interactivity as outlined in the Proposal section above So instead of just an “interactive video”, we will have “interactive content”, which will be a very powerful learning tool for the user.

2. We will allow for users to add interactivity to a video in the uploading process itself.

3. We will allow for different question types: Match the Following, Multiple Answer Questions, and so on.

4. We will allow creators to make content interactive within a specific context (e.g.: collection). For example, add a question set interception in a video only within a specific course. The video will be a normal one when consumed as an independent resource.

Localization Requirements

This task does not have any special localization requirements. The level of localization supported will be the same as the level of localization of the underlying tools: video player, QuML player, and Video Creation.

Telemetry Requirements

We will collect click level interaction when an interactive video is played. Interactive Videos will emit all telemetry that is already emitted by the included components: the video player, and the QuML player. In addition to this, some custom events specific to the behaviour of interactive videos will be defined.

Key Metrics

The main metric to track here will be engagement with the interactive video. Some of the metrics we can use for this purpose are listed below

The goal will be to understand which interactive videos are working better, which are not doing so well, and in general, understand what makes for a good interactive video. For example:

1. What is the sweet spot of interception point for driving engagement, i.e. do questions work better in the first 1 minute or after 3 minutes or after 5 minutes?

2. How long should my video be: 3-5 minutes or 8-10 minutes? How does this change depending on subject, age group, class, etc?

For example, NROER / NCERT has many videos which are 45 minute long (recorded lectures). I want to know how many pieces should I make of them to make them of ideal length (available from 2) and where to put the questions (available from 1).